Research map / Large language models / Retrieval and knowledge
ISEEQ: Information Seeking Question Generation Using Dynamic Meta-Information Retrieval and Knowledge Graphs
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use ISEEQ-ERL instead of T5-FT CANARD with triples | ISQ generation · QAD test | Rouge-L: 0.51 -> 0.67 | Helped |
| Use ISEEQ-ERL instead of T5-FT CANARD with triples | ISQ generation · QAMR test | BERTScore: 0.66 -> 0.77 | Helped |
| Use ISEEQ-ERL instead of T5-FT CANARD with triples | ISQ generation · FBC test | logical coherence (%): 68.2 -> 79.4 | Helped |
| Concatenate knowledge-augmented query encodings to passage encodings (ablation) | ISEEQ-ERL · QAD test | logical coherence (%): 24.6 -> 25.7 | Helped |
| Concatenate knowledge-augmented query encodings to passage encodings (ablation) | ISEEQ-ERL · QAMR test | logical coherence (%): 36.3 -> 37.0 | Helped |
| Concatenate knowledge-augmented query encodings to passage encodings (ablation) | ISEEQ-ERL · FBC test | logical coherence (%): 78.2 -> 79.4 | Helped |
| Use knowledge-enriched query embeddings instead of entity-only embeddings for KPR (ablation) | KPR · QAMR Wikinews | MAP: 0.35 -> 0.38 | Helped |
| Add entailment constraints and reverse cross-entropy for conceptual flow (ablation) | ISEEQ-RL · QAD test | logical coherence (%): 20.0 -> 22.0 | Helped |
| Replace single-passage fine-tuned question generation with dynamic multi-passage retrieval and ISEEQ generation | ISQ generation · QAD | Rouge-L: 20-35% | Helped |
| Train ISEEQ-ERL on QAD and test on QAMR | ISEEQ-ERL · QAMR test | logical coherence (%): 33.1 | Mixed |
| Use ISEEQ-ERL rather than T5-FT CANARD for human-rated questions | ISQ generation · Human evaluation, G1 diversity | Likert rating: 3.518 -> 3.759 | Helped |
| Use ISEEQ-ERL rather than T5-FT CANARD for human-rated questions | ISQ generation · Human evaluation, G2 coherence | Likert rating: 3.503 -> 3.843 | Helped |
Closest papers by meaning
- IQA-EVAL: Automatic Evaluation of Human-Model Interactive Question Answering (NeurIPS 2024 · similarity 0.86)
- Response Enhanced Semi-supervised Dialogue Query Generation (AAAI 2024 · similarity 0.85)
- WideSearch: Benchmarking Agentic Broad Info-Seeking (ICLR 2026 · similarity 0.85)
- UniGen: A Unified Generative Framework for Retrieval and Question Answering with Large Language Models (AAAI 2024 · similarity 0.85)
- From Retrieval to Generation: A Simple and Unified Generative Model for End-to-End Task-Oriented Dialogue (AAAI 2024 · similarity 0.85)
- QGen Studio: An Adaptive Question-Answer Generation, Training and Evaluation Platform (AAAI 2025 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.